summaryrefslogtreecommitdiffstats
path: root/third_party/aom/aom_dsp/x86/subtract_avx2.c
blob: 4389d123db6d6ad43b1168ceaebbd33a480ecb91 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
/*
 * Copyright (c) 2018, Alliance for Open Media. All rights reserved
 *
 * This source code is subject to the terms of the BSD 2 Clause License and
 * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
 * was not distributed with this source code in the LICENSE file, you can
 * obtain it at www.aomedia.org/license/software. If the Alliance for Open
 * Media Patent License 1.0 was not distributed with this source code in the
 * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
 */
#include <immintrin.h>

#include "config/aom_dsp_rtcd.h"

static INLINE void subtract32_avx2(int16_t *diff_ptr, const uint8_t *src_ptr,
                                   const uint8_t *pred_ptr) {
  __m256i s = _mm256_lddqu_si256((__m256i *)(src_ptr));
  __m256i p = _mm256_lddqu_si256((__m256i *)(pred_ptr));
  __m256i s_0 = _mm256_cvtepu8_epi16(_mm256_castsi256_si128(s));
  __m256i s_1 = _mm256_cvtepu8_epi16(_mm256_extracti128_si256(s, 1));
  __m256i p_0 = _mm256_cvtepu8_epi16(_mm256_castsi256_si128(p));
  __m256i p_1 = _mm256_cvtepu8_epi16(_mm256_extracti128_si256(p, 1));
  const __m256i d_0 = _mm256_sub_epi16(s_0, p_0);
  const __m256i d_1 = _mm256_sub_epi16(s_1, p_1);
  _mm256_store_si256((__m256i *)(diff_ptr), d_0);
  _mm256_store_si256((__m256i *)(diff_ptr + 16), d_1);
}

static INLINE void aom_subtract_block_16xn_avx2(
    int rows, int16_t *diff_ptr, ptrdiff_t diff_stride, const uint8_t *src_ptr,
    ptrdiff_t src_stride, const uint8_t *pred_ptr, ptrdiff_t pred_stride) {
  for (int32_t j = 0; j < rows; ++j) {
    __m128i s = _mm_lddqu_si128((__m128i *)(src_ptr));
    __m128i p = _mm_lddqu_si128((__m128i *)(pred_ptr));
    __m256i s_0 = _mm256_cvtepu8_epi16(s);
    __m256i p_0 = _mm256_cvtepu8_epi16(p);
    const __m256i d_0 = _mm256_sub_epi16(s_0, p_0);
    _mm256_store_si256((__m256i *)(diff_ptr), d_0);
    src_ptr += src_stride;
    pred_ptr += pred_stride;
    diff_ptr += diff_stride;
  }
}

static INLINE void aom_subtract_block_32xn_avx2(
    int rows, int16_t *diff_ptr, ptrdiff_t diff_stride, const uint8_t *src_ptr,
    ptrdiff_t src_stride, const uint8_t *pred_ptr, ptrdiff_t pred_stride) {
  for (int32_t j = 0; j < rows; ++j) {
    subtract32_avx2(diff_ptr, src_ptr, pred_ptr);
    src_ptr += src_stride;
    pred_ptr += pred_stride;
    diff_ptr += diff_stride;
  }
}

static INLINE void aom_subtract_block_64xn_avx2(
    int rows, int16_t *diff_ptr, ptrdiff_t diff_stride, const uint8_t *src_ptr,
    ptrdiff_t src_stride, const uint8_t *pred_ptr, ptrdiff_t pred_stride) {
  for (int32_t j = 0; j < rows; ++j) {
    subtract32_avx2(diff_ptr, src_ptr, pred_ptr);
    subtract32_avx2(diff_ptr + 32, src_ptr + 32, pred_ptr + 32);
    src_ptr += src_stride;
    pred_ptr += pred_stride;
    diff_ptr += diff_stride;
  }
}

static INLINE void aom_subtract_block_128xn_avx2(
    int rows, int16_t *diff_ptr, ptrdiff_t diff_stride, const uint8_t *src_ptr,
    ptrdiff_t src_stride, const uint8_t *pred_ptr, ptrdiff_t pred_stride) {
  for (int32_t j = 0; j < rows; ++j) {
    subtract32_avx2(diff_ptr, src_ptr, pred_ptr);
    subtract32_avx2(diff_ptr + 32, src_ptr + 32, pred_ptr + 32);
    subtract32_avx2(diff_ptr + 64, src_ptr + 64, pred_ptr + 64);
    subtract32_avx2(diff_ptr + 96, src_ptr + 96, pred_ptr + 96);
    src_ptr += src_stride;
    pred_ptr += pred_stride;
    diff_ptr += diff_stride;
  }
}

void aom_subtract_block_avx2(int rows, int cols, int16_t *diff_ptr,
                             ptrdiff_t diff_stride, const uint8_t *src_ptr,
                             ptrdiff_t src_stride, const uint8_t *pred_ptr,
                             ptrdiff_t pred_stride) {
  switch (cols) {
    case 16:
      aom_subtract_block_16xn_avx2(rows, diff_ptr, diff_stride, src_ptr,
                                   src_stride, pred_ptr, pred_stride);
      break;
    case 32:
      aom_subtract_block_32xn_avx2(rows, diff_ptr, diff_stride, src_ptr,
                                   src_stride, pred_ptr, pred_stride);
      break;
    case 64:
      aom_subtract_block_64xn_avx2(rows, diff_ptr, diff_stride, src_ptr,
                                   src_stride, pred_ptr, pred_stride);
      break;
    case 128:
      aom_subtract_block_128xn_avx2(rows, diff_ptr, diff_stride, src_ptr,
                                    src_stride, pred_ptr, pred_stride);
      break;
    default:
      aom_subtract_block_sse2(rows, cols, diff_ptr, diff_stride, src_ptr,
                              src_stride, pred_ptr, pred_stride);
      break;
  }
}